gh-60107: Remove a copy from RawIOBase.read (#141532)

If the underlying I/O class keeps a reference to the memory, raise BufferError.

Co-authored-by: Victor Stinner <[email protected]>

Author: cmaloney

Include/internal/pycore_global_objects_fini_generated.h

@@ -793,6 +793,7 @@ struct _Py_global_strings {
         STRUCT_FOR_ID(symmetric_difference_update)
         STRUCT_FOR_ID(tabsize)
         STRUCT_FOR_ID(tag)
+        STRUCT_FOR_ID(take_bytes)
         STRUCT_FOR_ID(target)
         STRUCT_FOR_ID(target_is_directory)
         STRUCT_FOR_ID(task)

Include/internal/pycore_global_strings.h

@@ -609,6 +609,25 @@ def readinto(self, b):
             with self.assertRaises(ValueError):
                 Misbehaved(bad_size).read()
 
+    def test_RawIOBase_read_gh60107(self):
+        # gh-60107: Ensure a "Raw I/O" which keeps a reference to the
+        # mutable memory doesn't allow making a mutable bytes.
+        class RawIOKeepsReference(self.MockRawIOWithoutRead):
+            def __init__(self, *args, **kwargs):
+                self.buf = None
+                super().__init__(*args, **kwargs)
+
+            def readinto(self, buf):
+                # buf is the bytearray so keeping a reference to it doesn't keep
+                # the memory alive; a memoryview does.
+                self.buf = memoryview(buf)
+                buf[0:4] = self._read_stack.pop()
+                return 3
+
+        with self.assertRaises(BufferError):
+            rawio = RawIOKeepsReference([b"1234"])
+            rawio.read(4)
+
     def test_types_have_dict(self):
         test = (
             self.IOBase(),

Include/internal/pycore_runtime_init_generated.h

@@ -0,0 +1,2 @@
+Remove a copy from :meth:`io.RawIOBase.read`. If the underlying I/O class
+keeps a reference to the mutable memory, raise a :exc:`BufferError`.

Include/internal/pycore_unicodeobject_generated.h

@@ -927,33 +927,33 @@ _io__RawIOBase_read_impl(PyObject *self, Py_ssize_t n)
         return PyObject_CallMethodNoArgs(self, &_Py_ID(readall));
     }
 
-    /* TODO: allocate a bytes object directly instead and manually construct
-       a writable memoryview pointing to it. */
     b = PyByteArray_FromStringAndSize(NULL, n);
-    if (b == NULL)
+    if (b == NULL) {
         return NULL;
+    }
 
     res = PyObject_CallMethodObjArgs(self, &_Py_ID(readinto), b, NULL);
     if (res == NULL || res == Py_None) {
-        Py_DECREF(b);
-        return res;
+        goto cleanup;
     }
 
     Py_ssize_t bytes_filled = PyNumber_AsSsize_t(res, PyExc_ValueError);
-    Py_DECREF(res);
+    Py_CLEAR(res);
     if (bytes_filled == -1 && PyErr_Occurred()) {
-        Py_DECREF(b);
-        return NULL;
+        goto cleanup;
     }
     if (bytes_filled < 0 || bytes_filled > n) {
-        Py_DECREF(b);
         PyErr_Format(PyExc_ValueError,
                      "readinto returned %zd outside buffer size %zd",
                      bytes_filled, n);
-        return NULL;
+        goto cleanup;
+    }
+    if (PyByteArray_Resize(b, bytes_filled) < 0) {
+        goto cleanup;
     }
+    res = PyObject_CallMethodNoArgs(b, &_Py_ID(take_bytes));
 
-    res = PyBytes_FromStringAndSize(PyByteArray_AsString(b), bytes_filled);
+cleanup:
     Py_DECREF(b);
     return res;
 }